Table 1.
Test items of the MABC-2 for Age Band 2 and 3.
Table 2.
An overview of the study samples, tasks, and methods.
Fig 1.
Distribution of the scores of the eight MABC-2 items, pilot study (n = 35).
The y-axis shows the number of participants achieving a given raw score (x-axis). Panels A-C, manual dexterity; D-E, aiming and catching; F-H, balance items. Good performance is represented on the right, and poor performance on the left of each x-axis. Ceiling effects are seen in a greater proportion of the distribution on the right of each histogram (e.g., panel G). Items where both hands (or legs) were scored are shown in the same panel, with the preferred or best limb in black and the non-preferred or other limb in grey.
Table 3.
Skewness, test of normality and chi square results of MABC-2 tasks for the pilot study.
Fig 2.
Distribution of the scores of the eight MABC-2 items, Experiment 1 (n = 161).
The y-axis shows the number of participants achieving a given raw score (x-axis). Panels A-C, manual dexterity; D-E, aiming and catching; F-H, balance items. Good performance is represented on the right, and poor performance on the left of each x-axis. Ceiling effects are seen in a greater proportion of the distribution on the right of each histogram (e.g., panel G). Items where both hands (or legs) were scored are shown in the same panel, with the preferred or best limb in black and the non-preferred or other limb in grey.
Table 4.
Skewness, test of normality and chi square results of MABC-2 tasks for the pilot study and Experiment 1.
Fig 3.
Distribution of the scores of the three MABC-2 items, Experiments 1 (n = 161) and 2 (n = 81) for best of two trials (black), average across trials (mid-grey) and Age Band 3 trial (light grey).
The y-axis shows the number of participants achieving a given raw score (x-axis). Good performance is reported on the right and poor performance on the left of each x-axis. Ceiling effects are seen in a greater proportion of the distribution on the right of each histogram (e.g., panel C). Items where both legs were scored are shown in the same panel, with the preferred or best limb in black and the non-preferred or other limb in grey. Three scoring methods are reported for each experiment. Age Band 3 data are reported in separate panels for Experiment 1 due to differences in number of participants.
Table 5.
Skewness, test of normality and chi square results of three MABC-2 tasks in Experiment 2, including an exploratory measure calculating errors.
Fig 4.
Algorithm for creating non-parametric ‘bootstrapped’ distributions of MABC-2 scores.
The three components of movement (Manual dexterity, MD; Aiming & catching, A&C; Balance, B), are represented in blues, reds, and greens respectively. Each component includes 2 or 3 items, shown in different shades. Each item includes between 2 and 10 trials, executed with the dominant or preferred hand or leg (P1, P2, … P10) or the non-dominant or non-preferred hand or leg (N1, N2). The full dataset of 388 children was used. On each of 300,000 iterations of the basic resampling procedure, a random child was selected. The dataset was checked for completeness, and for the child not having a DCD diagnosis. From that child’s full dataset, a random dataset was created, by sampling, within each item and each hand or leg randomly with replacement from the 10 (A&C) or 2 (all other items) available trials. From this random dataset, the ‘best’ and ‘mean’ scores for each item, hand, and leg were calculated, and the process repeated for another randomly selected child. After 300,000 random datasets had been created, the scores were ranked, converted into proportions, and converted to logit scores. These logit scores were used to create component standard scores and a total standard score, and these standard scores were converted into percentiles.
Fig 5.
Distributions of component standard scores and total standard scores following standard parametric (filled columns) and alternative non-parametric resampling methods (unfilled columns).
Component and total score percentiles were converted into logit scores and plot in bins of 0.25 width across all participants, normalising to a total plot height of 1. Note that the y-axis scale has been truncated for the non-parametric scores in the balance component. Just under 50% of all these scores were in a single bin.
Fig 6.
Relationship between parametric (x-axis) and non-parametric (y-axis) methods of assigning percentiles to scores for 225 8–10 year-old children performing the MABC-2.
Data points above the diagonal lines indicate higher percentiles were assigned for non-parametric than parametric; below the lines indicates higher for parametric than non-parametric.